The Process Map

Doug Sanders Bill Ross Jim Coleman

The Process Map

By Doug Sanders, Bill Ross, and Jim Coleman

Abstract

This paper describes a modification of Process Flowcharts and Cause and Effect Diagrams that are used
in conjunction with other tools and techniques to facilitate and document process investigation and
improvement.

Introduction

The pressure to continuously improve an organization’s products and services requires management
behavior and engineering decisions that reflect a knowledge of process and product performance as
never before. Understanding and managing the causal relationship between process variables and
product performance is not only desirable; it is a competitive necessity.

Process study might then be defined as the acquisition of knowledge about process parameters in order
to be able to manipulate process outputs in a predictable fashion with minimum variation (1 - The Analytic
Examination Of Time Dependent Variance Components). This work will necessarily include investigation
of potential sources of variation which may be present

Page 1 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

in a process in order to understand their impact on process outputs or product performance. Statistical
Process Control (SPC) and Design of Experiments (DOE) are powerful techniques to assist in rapidly
acquiring this knowledge. Of course, process and engineering knowledge provide the basis for these
tools.

Tools used in process study include Cause and Effect (C/E)

Diagrams and Flowcharts. The C/E Diagram provides assistance in the identification and prioritization of
potential sources of variation for investigation (possibly with Control Charts using various rational
subgrouping plans and DOEs). The Flowchart is a graphical tool used to helpsuppliers, process owners,
operators, engineers and customers understand how the steps in a process function together to deliver a
product or service. Once created, a flowchart is an invaluable tool for communicating, training, assisting in
the identification of value-added and nonvalue-added steps, and analyzing of product flow issues. The
Team Handbook discusses four commonly used flowcharts and the C/E Diagram (2 -

The Team Handbook, Peter R. Scholtes, Joiner Associates).

Flowcharts facilitate process investigation, but typically do not provide insight into the
mechanisms driving process levels and variability. On the other hand, the potential causal
structure captured in the C/E Diagram is not tied to location in the process,

Page 2 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

nor is the current state of process knowledge depicted. Seldom is the full potential of the combination of
Process Flowcharts and C/E

Diagrams realized. Neither of these two tools indicate whether the potential sources of variation are
currently being managed or if they can be managed at all. A further drawback lies in the fact that they

typically do not become living documents, updated as new knowledge is acquired concerning the causal
structure of the process. The shortcomings of the C/E Diagram and process flowchart potentially inhibit
the investigation and management of the causal structure.

The Process Map is a tool that displays current process knowledge and is a supplement to many of the
traditional process investigation tools. It enhances the usual flowcharts with the type of knowledge
captured in C/E Diagrams. The apriori construction of the Process Map can dramatically increase the
effectiveness of statistical techniques by facilitating the critical thinking required to gain and utilize an
understanding of the relationship between process variables and product characteristics. As a working
document, the Process Map is used to continually capture the existing state of process knowledge and
the means of management of that knowledge.

Terminology

Page 3 of 18
The Process Map
 Doug Sanders Bill Ross Jim Coleman

The intent of process study is to develop a deeper understanding of the transformation of process inputs
and variables (x’s) into end- product characteristics (Y’s), or in matrix notation, Y = f(x).

At each step in the process, the transformation of the process parameters (x’s) to end-product
characteristics may be monitored by in-process outputs (y’s). When the choice is made to monitor in-
process outputs, it is usually to make sure that any problems with the product are detected and corrected
before substantial time and money have been invested. In reality, these in-process outputs are a function
of the process parameters. In other words, y = f (x) and

Y = g (y) = f (x).

Further insight into these functions can be gained by classifying each of the x’s into one of two categories:
controllable parameters and noise parameters. If a process parameter is set at a certain value and
maintained within a particular range it is considered controllable. Examples of controllable process
parameters include the following: feed and speed for a machining process, oven temperature and cure
time in a gluing operation, gas pressure and purity in a plasma cleaning operation, the number of tellers
working per shift at a bank, or the time a burger is allowed to remain under a heat lamp at a restaurant
before disposal. In each instance, the process parameter can be set and maintained around

Page 4 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

a desired value. A parameter that can not be, or is preferably not, set and maintained around a desired
value due to cost, physical, or other constraints is considered noise. Examples of noise parameters
include the following: ambient temperature for a machining process, relative humidity in a gluing
operation, the number and complexity of transactions per customer at a bank, or the time between
purchase and consumption of a burger at a restaurant. In each instance, the parameter is difficult or
costly to control.

It should be noted that a particular factor may be treated differently across processes. For instance,
ambient temperature and humidity might initially be considered noise parameters in a gluing operation,
while these same x’s are controllable parameters for the production of semiconductors in a clean room
environment. Of course, the decision to control a parameter for a particular process might be reevaluated
and could change as knowledge concerning the impact of variation in that parameter on end-product
characteristics is gained.

The classification of a process parameter as either controllable or noise does not necessarily imply
anything about the parameter’s impact on end-product characteristics (Y’s). Typical variation in noise
parameters can have a substantial impact on the product’s

Page 5 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

performance. Likewise, variability in a controllable process parameter across certain levels may have little
or no impact on the product’s performance. Insight into the relationship between process parameters and
product performance; Y = f(x), can not be gained by simply categorizing the x’s as controllable or noise.
This relationship can only be understood through experimentation or prolonged observation using SPC.
This categorization, however, does help to illustrate the current state of process management and can be
used to enhance the design and analysis of studies utilizing DOE and SPC.

If process investigation reveals that variation in a controllable parameter or a noise parameter has a
significant impact on product performance, that parameter is considered a critical x. Only through a
combination of engineering and process knowledge, supplemented by process investigation can these
critical x’s be discovered and confirmed. Once a critical x is identified, steps must be taken to ensure that
it is managed to the point that it will not cause undue variation in the product’s performance. Such steps
are often called Standard Operations. Until the product design or manufacturing process can be made
robust to variation in these critical x’s, Standard

Page 6 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

Operations are used to manage the x’s within a certain range so that end-product variability is minimized.

To reiterate and expand, the intent of process study is to develop a deeper understanding of the
transformation of process inputs and variables (x’s) to end-product characteristics (y’s). These inputs and
variables (process parameters) may or may not be currently managed. If they are being managed they
are considered controllable (C), if they are not managed (due to cost or other constraints) they are noise
(N). When the impact of variation in process parameters is empirically validated as influential in terms of
variation in end-product characteristics (possibly through a series of DOEs), the process parameter is
labeled critical. A simple example for a manual drilling process is shown in Figure 1.

Page 7 of 18
The Process Map
Doug Sanders Bill Ross Jim Coleman
y = Part Stability on Stand Part Flatness on Stand y = Part Position on Stand
Y = Hole Diameter
Hole Concentricity Hole Taper
Align
Clamp Part on Drill Stand
Part onto Drill Stand
Feed Rate Clamp Force
Speed Clamp Location
Coolant Type Tool Design Tool Age Material Hardness
Part Stability on Stand Part Flatness on Stand
Figure 1: Drilling Process Map
Page 8 of 18
Drill Hole
(N) (N) (C) (N)
(C) Part Cleanliness Stand Cleanliness
(C) (C) Age of Locating Pins Cleanliness of Locating Pins
(N) (C) (N)
(C) = Controllable (N) = Noise
* = Critical
* (C) (N)
*
*
Legend
Construction of the Process Map
To construct a meaningful Process Map, careful consideration must
be given to the scope of the process. What are the process
boundaries and expected outcomes (Y’s)? What are the target
values for product characteristics, and what is the penalty for
deviation from these target values? How will management know if
the process is performing in a manner that enhances the
organization’s competitive stance? Which Y’s are truly important to
the customer? While the answers to these questions are process
dependent, consideration is required before beginning process
investigation. The scope of the Process Map depends on these
answers.
The Process Map

Doug Sanders Bill Ross Jim Coleman

Process Maps are hierarchical in nature. A macro Process Map describes the major activities (e.g.
assembly of sub-assemblies) needed to complete and deliver a product or service to the external
customer. It is possible and often necessary to divide a high-level map into detailed sub maps (e.g.
assembly of components into a sub-component assembly). Detailed sub-maps may require further
sub-division into specific operations or require expansion upstream in the product flow (e.g. work on the
sub-component assembly process might lead to the component fabrication process).

Frequently Process Maps cross various functional areas in an organization. After determining the scope
and the level of the Process Map, the input parameters to the process are determined. These input
parameters are typically those things that initiate the process. Examples include raw material and raw
material characteristics, a partially completed product (sub-assemblies and components), a work order or
customer request, safety and government regulations, and tools. This information is captured and
displayed in a simple graphical manner, as illustrated in Figure 2.

Page 9 of 18
The Process Map
Doug Sanders Bill Ross Jim Coleman
Inputs
Cylinder
Outputs
Machine Operating Guide
Cleaning & Milling
Cylinder Head Flatness Operator Instructions (OI) Cylinder Head Hardness Cylinder Head Cleanliness Cleaning
Fluid Cutting Tool
Figure 2 - Input/Output Diagram for a simple machining operation
A flowchart depicting all the process steps, including inspection
and rework, is constructed next. It is extremely important that the
process be mapped in the as-is condition. This requires
interrogation of the process over various sampling intervals,
including multiple operators, multiple shifts, and changing noise
conditions. Each formal or informal inspection yields in-process
outputs (y’s) by which the process may be monitored. Keep in
mind, data on these parameters is often not recorded or even
measured. Frequently y’s must be identified with engineering
knowledge rather than processing experience.
The Process Flowchart - illustrating all of the steps in the process,
the end-product characteristics (Y’s), and the in-process outputs
(y’s) - is now prepared for the addition of process parameters (x’s).
Page 10 of 18
The Process Map
 Doug Sanders Bill Ross Jim Coleman

All potential process parameters should be identified and categorized as controllable (C), or as noise (N).

Categorizing the x’s as controllable or noise helps us to understand how the parameters are currently
being managed. This illustrates the current state of process belief; that is, how variation in the process
parameters translates into variation in the process outputs. Early versions of the Process Map will
typically be based on scientific theory, engineering knowledge, and operator experience.

Through process investigation and statistical techniques including DOE and SPC, this list of all process
parameters can be filtered so that critical process parameters are identified. The variation in these critical
parameters has a significant impact on end-product characteristics, Y’s, or on in-process outputs, y’s, that
in turn effect the end-product. For example, the number of tellers working during a shift at a bank, x, may
have a significant impact on the time a customer is required to wait in line prior to service, y, which in turn
may have a dramatic impact on the customer’s satisfaction with the banks service, Y.

An Illustrative Example

Through process investigation, a manufacturer of compressors has learned that the flatness of the
cylinder head is critical to product performance. An experiment was designed to determine which

Page 11 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

process parameters, or x’s, have a critical effect on flatness. The experimenter, focusing on the milling
operation, generated a design to examine the effect of mill feed rate, depth of cut and tool design on
flatness. By constructing the initial Process Map shown in

Figure 3, the experimenter learned several important things prior to running the design. First, the
experimenter learned that each piece is rigorously cleaned prior to milling. The criticality of this cleaning
operation was unknown. Additionally, the experimenter learned

that each of the three operators had developed his/her own fixture for holding the part during milling.
While the design of the fixture may have a critical impact on the flatness of the part, nothing was in place
to control which fixture was used. For this reason, fixture design was considered a noise parameter.
Lastly, the experimenter learned of two other noise parameters that may impact flatness. The allowable
amount of tool wear and the hardness of the cylinder head were not currently being controlled.

Page 12 of 18
The Process Map
Doug Sanders Bill Ross Jim Coleman
y = Head Stability in Fixture Head Orientation in Fixturey = Head Cleanliness
Y = Head Flatness
Place Cylinder Head in Fixture
Fixture Pressure Fixture Design
Mill Feed Rate Depth of Cut Mill Tool Design Mill Tool Wear Material Hardness
Head Stability
Figure 3: Compressor Cylinder Head Initial Process Map
(Before Experimentation)
Page 13 of 18
Mill Cylinder Head Surface
(C) (C) (N)
Clean Cylinder Head
Cleaning Time Cleaning Fluid Cleaning Method
(C) = Controllable (N) = Noise
* = Critical
(C) (N)
(C) (C)
Head Cleanliness
(C) (N) (N) Legend
Information from the Process Map led to an experiment designed to
investigate the following factors: Cleaning Time, Cleaning Fluid,
Fixture Pressure, Mill Feed Rate, Depth of Cut, and Tool Design.
While the hardness of the material, the amount of tool wear, and
the design of the clamping fixture were not normally controlled, the
experimenter could control these factors for the duration of the
study and decided to include them in the design. The results for
the 16 run design are easily interpreted using the effect chart
illustrated in Figure 4.
The Process Map
Doug Sanders Bill Ross Jim Coleman
Fixture Pressure (-)
Fixture Design ) snorcim(s sentalF30(-)
25Cleaning 20Time
15(+)
(+)
Page 14 of 18
Tool Age* Mill Feed Rate
(+)
10 (-) 5 Tool Age0

* Cleaning Time
Tool Age* Depth of Cut
(+)
(-)
Figure 4 - Experimental Results from machining DOE
Tool AgeDepth of Cut
Mill Feed
(-) Rate
(+)
(-)
(-)
(+)
* Cleaning
(-)
(-)
Fluid (+)
(+) Cleaning
Fluid
(-)
(+)
(+)
(-)
Tool Age (+)
(-)(-)
(-)
(+)
(+)
(-)
(+) Tool
Tool Design
(+) Material Hardness
Age* Fixture Design
Tool Age*Tool Design

On the effect chart in Figure 4, the magnitude of each factor and interaction relative to the average
flatness in the experiment is shown. Note that only 6 of the two-way interactions are illustrated
on the chart because of the resulting confounding for this particular design. For example, the interaction
between tool age and depth of cut (Tool Age * Depth of Cut) is confounded with the
interaction between Material Harness and Fixture Design (Material Hardness * Fixture Design). Because
the effect of these two interactions will be identical, only one of the two is illustrated. As with all fractional
factorial experimental designs, the confounding must be carefully considered prior to selecting the design
and when interpreting experimental results.

The Process Map

Doug Sanders Bill Ross Jim Coleman

For this particular experiment, the average flatness was 14.8 microns. The magnitude and direction of the
effect for any given factor or interaction can be calculated by examining the difference between the
average flatness at the high level of the factor and the average flatness at the low level of the factor. For
example, consider the effect of fixture pressure. In this experiment, the two levels of fixture pressure were
100 PSI (-) and 200 PSI (+). The average flatness when the fixture pressure was held at the (+) level was
4.6 microns. When the fixture pressure was held at the (-) level, however, the average flatness was 25.0
microns. Hence, changing fixture pressure from 100 PSI to 200 PSI reduced the average cylinder head
flatness by 20.4 microns (4.6 – 25.0 = -20.4). Because the cylinder head is responsible for sealing off the
chamber it is desirable to set up the process so flatness is minimized. Thus for right now, fixture pressure
should be set to 200 PSI, and further experimentation should be performed on this and other important
factors to determine optimal settings. The other factors and interactions can be analyzed in a similar
manner.

The effect chart can also be used to compare the relative importance of the factors in the experiment.
Note that the difference between the average flatness at each level of a given factor or interaction is
represented by the length of the vertical line drawn between the two levels.

Page 15 of 18
The Process Map

Doug Sanders Bill Ross Jim Coleman

As can be seen from Figure 4, the experimenter learned several valuable things that would not have
shown up in the initial design. Cleaning time and cleaning fluid seem to have little impact on the flatness
of the part, as illustrated by the relatively short lines. Further experimentation revealed that the parts did
not have to be cleaned at all. This step in the process was eliminated, reducing the cycle time and labor
required to fabricate the part. The results also indicated that the factors that were included in the initial
design, across typical levels, have little to do with the flatness of the part. The important factors were
those that were added to the experiment after constructing the Process Map. As can be seen from Figure
4, critical process parameters included the design of the holding fixture, the fixture pressure and the
hardness of the incoming cylinder heads. Through further experimentation, the optimal fixture design and
pressure were determined and are currently being controlled. However, the process owner is uncertain
how to control the hardness of the incoming cylinder heads. While this process parameter is critical, it is
still considered a noise parameter because it is not currently being controlled. Note that the Process Map
has been updated to reflect the new level of process understanding (see Fig. 5). Critical process
parameters have been identified, and the current state of parameter control is shown.

Page 16 of 18
The Process Map
Doug Sanders Bill Ross Jim Coleman
y = Head Stability in Fixture
Head Orientation in Fixture
Mill Feed Rate Depth of Cut Mill Tool Design Mill Tool Wear Material Hardness
Head Stability Head Orientation
Page 17 of 18
Y = Head Flatness
Place Cylinder Head in Fixture
Mill Cylinder Head Surface
* (C)
Fixture Pressure
(C) * (C)
Fixture Design
(C) (C)
Legend
(N) *
(N)
(C) = Controllable (N) = Noise
* = Critical
Figure 5: Compressor Cylinder Head Process Map (After Experimentation)
Conclusion
Graphically combining the knowledge typically depicted on a
flowchart with that from a Cause and Effect Diagram, the Process
Map overcomes the weaknesses of the two tools used
independently. Additionally, the Process Map provides a clear
understanding of the current state of process management by
classifying each parameter as controllable or noise. As knowledge
is gained through prolonged observation using SPC or through a
series of experiments, the Process Map is updated to highlight
critical process parameters. Through this living document, the
current state of process knowledge is readily available to all
interested parties, greatly enhancing classical process improvement
The Process Map

Doug Sanders Bill Ross Jim Coleman

techniques including SPC and DOE. By using this tool to understand and manage the causal relationship
between process parameters and product performance, any process can be continuously improved to
ensure success in today’s competitive environment.

References

1. Sanders, R., Leitnaker, M., and Sanders, D. (1994-1995). “The Analytic Examination of
Time-Dependent Variance Components”. Quality Engineering, Vol. 7, No. 4, pp. 315-336.

2. Scholtes, P. R. (1988). The Team Handbook. Joiner Associates.

 Madison, WI.

Key Words and Phrases

Process Map, Flowchart, Cause and Effect Diagram, Fishbone

Diagram, Ishikawa Diagram, Process Improvement, process

investigation, Statistical Process Control (SPC), Design of

Experiments (DOE), Components of Variation.

Page 18 of 18